Harshit Arora

@UTDIMole Kroos

Guess the player Very hard.

@kxie06 Dmed you!

We’re scaling rapidly and hiring for full-time roles in SF. This is a rare chance to work directly with me and an incredible team to build the future of enterprise AI. Our belief is that every company will eventually run on private, custom AI tailored to its workflows. The challenge is that building high-quality custom systems today still looks too much like consulting. Our vision is to scale services with software margins. We’re building the infrastructure to make that possible, and we need exceptional people to do it. We’re looking for someone who: - learns extremely fast - is passionate about their work and takes ownership - loves talking to customers and naturally builds strong relationships I don’t care what your y-intercept is, only your slope. DM me if interested.

@DeadlineDayLive @grok summarize a bit on how this works ?

🚨 And now Burnley will earn more than THREE TIMES the amount Inter received for winning Serie A just by getting relegated from the Premier League. 😭💰 European football has completely lost the plot. 🌍

@robin_codes @robin_codes unable to dm you

I’m hiring across several roles at Artie: - Product Engineer - Software Engineer - Solutions Engineer We’re growing quickly and looking for people who are not only exceptional engineers, but also a genuine cultural fit. We’re 5 days a week in person in San Francisco, and we sponsor visas. Links to apply are in the thread. Referrals are always welcome.

@ignisjuliuxo Unable to DM you

Hiring funding engineers. DM open! peerlist.io/company/auriko…

Cannot believe it

🚨 𝐎𝐅𝐅𝐈𝐂𝐈𝐀𝐋: 𝐀𝐑𝐒𝐄𝐍𝐀𝐋 𝐀𝐑𝐄 𝐂𝐇𝐀𝐌𝐏𝐈𝐎𝐍𝐒 𝐎𝐅 𝐄𝐍𝐆𝐋𝐀𝐍𝐃 𝐅𝐎𝐑 𝐓𝐇𝐄 𝐅𝐈𝐑𝐒𝐓 𝐓𝐈𝐌𝐄 𝐒𝐈𝐍𝐂𝐄 𝟐𝟎𝟎𝟒 🏆

@JustFreshKicks @grok is this official drop?

ICEMAN for FC Barcelona 🧊

@djm5226 @GlitchedDeals They canceled my order

@GlitchedDeals Good looks

LEGO’S ARE GLITCHING ON KOHL’S INSTACART Search Kohl’s on Instacart and check all toys Up to 80% off toys like Legos and other brands This glitch varies per location so try checking all stores

@GlitchedDeals Leggoooo ty sir 🫡

@TheClubJunto @grok verify this claim

Foreign Capital May Not Return Soon: Protect Your Downside 1. 85% of India's capital outflows going to: Korea, Taiwan, China 2. Their 2026 Forward Earnings: 2X/3X of India 3. Their AI/Semicon Growth Supercycle vs. India’s IT Slump 4. Rupee fall erodes FII/FDI capital DATA: Where Is the Money Going? a. Over ₹1 lakh crore FII capital exited India in 2026 (YTD). b. 60% to Korea & Taiwan: AI/Semiconductor supply chain stocks c. 25% to China: Bottom-fishing in undervalued tech stocks d. 15% to “Safe Havens”: U.S. Treasuries and Gold 2025 Actual Earnings Growth India (Nifty 50): 8.1% Korea (Kospi): 76% Taiwan (TWSE): 27% China (CSI 300): 12% 2026 Earnings Growth (Est.) India (Nifty 50): 9% Korea (Kospi): 130% Taiwan (TWSE): 22% China (CSI 300): 13% NOTE: Estimates are from Goldman Sachs, JP Morgan, Morgan Stanley, Bernstein reports. 2026 Forward P/E Comparison India (Nifty 50): 18.1x Korea (Kospi): 8.8x Taiwan (TWSE): 19.7x China (CSI 300): 13.9x NOTES: a. Even after the recent sell-off, India still remains one of the most expensive markets in Asia. b. Korea has seen an earnings explosion in memory chips, but that market still offers deep value to investors at 8.8x forward P/E. c. Taiwan is getting expensive, but AI/Semiconductor dominance of Taiwanese companies and TSMC’s pricing power is still a great attraction. Estimated earnings growth of Taiwan in 2026 is 2.5x of India. d. China is no longer at “distressed” price levels of 2024, but it is still fundamentally 25% cheaper than India. P/B Ratios March 2026 India (Nifty 50): 3.14x Korea (Kospi): 0.90x Taiwan (TWSE): 2.2x China (CSI 300): 1.45x NOTE: India is the world’s most expensive major market on P/B ratio basis. India’s price-to-book is 1.5x of Taiwan, 2.5x of China, and 3.5x of Korea. In other words, it is 3.5 times more expensive than Korea for every unit of net asset value as of March 2026. KOREA a. Landmark domestic corporate governance reforms combined with a global semiconductor growth supercycle have created a rare “double-alpha” opportunity for investors. b. A stunning 130% earnings growth projected for KOSPI in 2026, led by Samsung and SK Hynix. c. Korea’s forward P/E of 8.8x is half the valuation of Indian large caps as of today. TAIWAN a. Investors are moving away from AI software/LLM builders to AI hardware (semiconductors and server infrastructure) where Taiwan dominates. b. TSMC alone controls 70% of the market share, with gross margins of 62%. c. Taiwan 50 Index with an estimated earnings growth of 22% at a PEG ratio of just 0.9x makes it twice as attractive as Indian large caps on a growth-adjusted basis. CHINA a. China’s industrial output in Jan-Feb, 2026 jumped by 6.3%, beating estimates by far. b. High-tech FDI increased by 20.4% following the government’s massive consumption stimulus package in late 2025. c. As of March 2026, MSCI China trades at a forward P/E 11.9x, representing 48% discount compared to MSCI India’s 23x. This 48% valuation gap is at a decade-high, leading to an FII pivot towards China. (Foreign investors use MSCI benchmarks.) INDIA a. With 85% dependency on oil imports, current account deficit (CAD) widening, rupee getting weaker, and no tech exports hedge, India presents an asymmetric risk (from the viewpoint of foreign investors.) b. China is energy-secure, while Korea & Taiwan’s high-margin tech exports effectively subsidize their increased oil import bills. c. IT service exports, which is India’s solitary global competitive edge, is getting threatened by AI. Legacy coding tasks are getting automated. ENDPIECE: Don’t Fight Mean Reversion Don’t believe vested interests whose careers are built on the thesis of “Stocks Only Go Up.” The world has shifted from the era of “Growth at Any Price” (GAP) to “Growth at a Reasonable Price” (GRP). That’s what has hit India. Now either India delivers exceptional earnings growth in 2026 to justify its P/E multiples, OR the AI growth bubble bursts worldwide. Until then, play defensive, avoid FOMO, and wait for the loose ball. Your time will come. @arabicatrader

@Pricerrors literally ordered and recieved in 6 hours

4K DASH CAM FOR $17 ON AMAZON Found a promo code that drops this $50 dash cam to $17 Comes with a 64GB SD card, WiFi app control, and parking mode Use code IWY4W6S2 at checkout pricedoffers.com/rqbxx #ad

@bytedunks @physical_int This is the kind of transparency robotics needs. Real costs, real constraints, real lessons. Respect for sharing the full stack.

To understand what it takes to build a humanoid robot with model-based control, we finetuned @physical_int 's (PI) Pi05 model for our custom use case and environment. We incurred ~$10K in hardware costs, compared to the typical ~$20K set up (DROID/ALOHA). Here are the lessons and challenges we faced building the first working prototype (shown in the video) in 3 months. Part 1: Hardware, Software, Model Selection, Custom Embodiment, Inference, Embedded Hardware, Hierarchical Planner Part 2: Model Evaluation, Data Collection, Model Training, Simulation and Teleoperation We hope sharing our experience accelerates the learning of others who are in a similar starting point.

@YuXiang_IRVL This looks awesome Yu, any idea how oclusions could be handled at scale with this?

Temporal and spatial alignment between glove and camera data. It is not an easy task, but it’s coming together. Human data for robots. #Robotics

@DrJimFan @ruijie_zheng12 Awesome stuff Jim would love to chat more about this!

This is a huge team work at NVIDIA Robotics. Check out @ruijie_zheng12's deep dive: - Website: research.nvidia.com/labs/gear/egos… - Paper: arxiv.org/abs/2602.16710 x.com/ruijie_zheng12…

We trained a humanoid with 22-DoF dexterous hands to assemble model cars, operate syringes, sort poker cards, fold/roll shirts, all learned primarily from 20,000+ hours of egocentric human video with no robot in the loop. Humans are the most scalable embodiment on the planet. We discovered a near-perfect log-linear scaling law (R² = 0.998) between human video volume and action prediction loss, and this loss directly predicts real-robot success rate. Humanoid robots will be the end game, because they are the practical form factor with minimal embodiment gap from humans. Call it the Bitter Lesson of robot hardware: the kinematic similarity lets us simply retarget human finger motion onto dexterous robot hand joints. No learned embeddings, no fancy transfer algorithms needed. Relative wrist motion + retargeted 22-DoF finger actions serve as a unified action space that carries through from pre-training to robot execution. Our recipe is called "EgoScale": - Pre-train GR00T N1.5 on 20K hours of human video, mid-train with only 4 hours (!) of robot play data with Sharpa hands. 54% gains over training from scratch across 5 highly dexterous tasks. - Most surprising result: a *single* teleop demo is sufficient to learn a never-before-seen task. Our recipe enables extreme data efficiency. - Although we pre-train in 22-DoF hand joint space, the policy transfers to a Unitree G1 with 7-DoF tri-finger hands. 30%+ gains over training on G1 data alone. The scalable path to robot dexterity was never more robots. It was always us. Deep dives in thread:

@ruijie_zheng12 @ruijie_zheng12 would love to chat further on this!

@ruijie_zheng12 Awesome stuff, Ruijie! Looks super interesting!

Proud to introduce EgoScale: We pretrained a GR00T VLA model on 20K+ hours of egocentric human video and discovered that robot dexterity can be scaled, not with more robots, but with more human data. A thread on 🧵what we learned. 👇

@stepjamUK Super interested Stephen, dm'ed you!

If you’re a robotics company and want to share what you’re building with the right audience, drop a comment below and lets continue to grow the community!

@stevepolacek Great read, Steve! I'm looking to set it up myself to optimize work and research.

@noah_schochet Dmed you!

If anyone will be in Austin for SXSW and wants to chat about giant construction robots, hit me up